Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
94.81% |
73 / 77 |
|
75.00% |
3 / 4 |
CRAP | |
0.00% |
0 / 1 |
| GrepPages | |
94.81% |
73 / 77 |
|
75.00% |
3 / 4 |
20.06 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
1 | |||
| init | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| execute | |
86.67% |
26 / 30 |
|
0.00% |
0 / 1 |
11.29 | |||
| findPages | |
100.00% |
31 / 31 |
|
100.00% |
1 / 1 |
7 | |||
| 1 | <?php |
| 2 | // phpcs:disable MediaWiki.Files.ClassMatchesFilename.NotMatch |
| 3 | use MediaWiki\Content\TextContent; |
| 4 | use MediaWiki\Language\Language; |
| 5 | use MediaWiki\Page\WikiPageFactory; |
| 6 | use MediaWiki\Revision\RevisionRecord; |
| 7 | use MediaWiki\Title\Title; |
| 8 | use MediaWiki\WikiMap\WikiMap; |
| 9 | use Wikimedia\Rdbms\IExpression; |
| 10 | use Wikimedia\Rdbms\LikeValue; |
| 11 | |
| 12 | // @codeCoverageIgnoreStart |
| 13 | require_once __DIR__ . '/Maintenance.php'; |
| 14 | // @codeCoverageIgnoreEnd |
| 15 | |
| 16 | /** |
| 17 | * Search pages for a given regex |
| 18 | * |
| 19 | * @ingroup Maintenance |
| 20 | */ |
| 21 | class GrepPages extends Maintenance { |
| 22 | /** @var Language */ |
| 23 | private $contLang; |
| 24 | |
| 25 | /** @var WikiPageFactory */ |
| 26 | private $wikiPageFactory; |
| 27 | |
| 28 | public function __construct() { |
| 29 | parent::__construct(); |
| 30 | $this->addDescription( 'Search the source text of pages for lines matching ' . |
| 31 | 'a given regex, and print the lines.' ); |
| 32 | $this->addOption( 'prefix', |
| 33 | 'Title prefix. Can be specified more than once. ' . |
| 34 | 'Use e.g. --prefix=Talk: to search an entire namespace.', |
| 35 | false, true, false, true ); |
| 36 | $this->addOption( 'show-wiki', 'Add the wiki ID to the output' ); |
| 37 | $this->addOption( 'pages-with-matches', |
| 38 | 'Suppress normal output; instead print the title of each page ' . |
| 39 | 'from which output would normally have been printed.', |
| 40 | false, false, 'l' ); |
| 41 | $this->addArg( 'regex', 'The regex to search for' ); |
| 42 | } |
| 43 | |
| 44 | private function init() { |
| 45 | $services = $this->getServiceContainer(); |
| 46 | $this->contLang = $services->getContentLanguage(); |
| 47 | $this->wikiPageFactory = $services->getWikiPageFactory(); |
| 48 | } |
| 49 | |
| 50 | public function execute() { |
| 51 | $this->init(); |
| 52 | |
| 53 | $showWiki = $this->getOption( 'show-wiki' ); |
| 54 | $wikiId = WikiMap::getCurrentWikiId(); |
| 55 | $prefix = $this->getOption( 'prefix' ); |
| 56 | $regex = $this->getArg( 0 ); |
| 57 | $titleOnly = $this->hasOption( 'pages-with-matches' ); |
| 58 | |
| 59 | if ( ( $regex[0] ?? '' ) === '/' ) { |
| 60 | $delimRegex = $regex; |
| 61 | } else { |
| 62 | $delimRegex = '{' . $regex . '}'; |
| 63 | } |
| 64 | |
| 65 | foreach ( $this->findPages( $prefix ) as $page ) { |
| 66 | $content = $page->getContent( RevisionRecord::RAW ); |
| 67 | $titleText = $page->getTitle()->getPrefixedDBkey(); |
| 68 | if ( !$content ) { |
| 69 | $this->error( "Page has no content: $titleText" ); |
| 70 | continue; |
| 71 | } |
| 72 | if ( !$content instanceof TextContent ) { |
| 73 | $this->error( "Page has a non-text content model: $titleText" ); |
| 74 | continue; |
| 75 | } |
| 76 | |
| 77 | $text = $content->getText(); |
| 78 | |
| 79 | if ( $titleOnly ) { |
| 80 | if ( preg_match( $delimRegex, $text ) ) { |
| 81 | if ( $showWiki ) { |
| 82 | echo "$wikiId\t$titleText\n"; |
| 83 | } else { |
| 84 | echo "$titleText\n"; |
| 85 | } |
| 86 | } |
| 87 | } else { |
| 88 | foreach ( StringUtils::explode( "\n", $text ) as $lineNum => $line ) { |
| 89 | $lineNum++; |
| 90 | if ( preg_match( $delimRegex, $line ) ) { |
| 91 | if ( $showWiki ) { |
| 92 | echo "$wikiId\t$titleText:$lineNum:$line\n"; |
| 93 | } else { |
| 94 | echo "$titleText:$lineNum:$line\n"; |
| 95 | } |
| 96 | } |
| 97 | } |
| 98 | } |
| 99 | } |
| 100 | } |
| 101 | |
| 102 | public function findPages( $prefixes = null ) { |
| 103 | $dbr = $this->getReplicaDB(); |
| 104 | $orConds = []; |
| 105 | if ( $prefixes !== null ) { |
| 106 | foreach ( $prefixes as $prefix ) { |
| 107 | $colonPos = strpos( $prefix, ':' ); |
| 108 | if ( $colonPos !== false ) { |
| 109 | $ns = $this->contLang->getNsIndex( substr( $prefix, 0, $colonPos ) ); |
| 110 | $prefixDBkey = substr( $prefix, $colonPos + 1 ); |
| 111 | } else { |
| 112 | $ns = NS_MAIN; |
| 113 | $prefixDBkey = $prefix; |
| 114 | } |
| 115 | $prefixExpr = $dbr->expr( 'page_namespace', '=', $ns ); |
| 116 | if ( $prefixDBkey !== '' ) { |
| 117 | $prefixExpr = $prefixExpr->and( |
| 118 | 'page_title', |
| 119 | IExpression::LIKE, |
| 120 | new LikeValue( $prefixDBkey, $dbr->anyString() ) |
| 121 | ); |
| 122 | } |
| 123 | $orConds[] = $prefixExpr; |
| 124 | } |
| 125 | } |
| 126 | $lastId = 0; |
| 127 | do { |
| 128 | $res = $dbr->newSelectQueryBuilder() |
| 129 | ->queryInfo( WikiPage::getQueryInfo() ) |
| 130 | ->where( $orConds ? $dbr->orExpr( $orConds ) : [] ) |
| 131 | ->andWhere( $dbr->expr( 'page_id', '>', $lastId ) ) |
| 132 | ->limit( 200 ) |
| 133 | ->caller( __METHOD__ ) |
| 134 | ->fetchResultSet(); |
| 135 | foreach ( $res as $row ) { |
| 136 | $title = Title::newFromRow( $row ); |
| 137 | yield $this->wikiPageFactory->newFromTitle( $title ); |
| 138 | $lastId = $row->page_id; |
| 139 | } |
| 140 | } while ( $res->numRows() ); |
| 141 | } |
| 142 | } |
| 143 | |
| 144 | // @codeCoverageIgnoreStart |
| 145 | $maintClass = GrepPages::class; |
| 146 | require_once RUN_MAINTENANCE_IF_MAIN; |
| 147 | // @codeCoverageIgnoreEnd |